Learning Speaker-specific Lip-to-Speech Generation | IEEE Conference Publication | IEEE Xplore